High IQ Society Requirements
Thresholds, Accepted Tests, and What Actually Gets You In

0 Quick Answer

Updated August 16, 2026 by Structural. High IQ societies admit members who reach a stated percentile on an approved, supervised test. Mensa, the largest, sets its threshold at the 98th percentile. Societies above it set progressively higher ones, up to thresholds so extreme that no standardised instrument can measure them.

Direct answer: two routes exist everywhere. Take the society's own supervised admission test, or submit prior evidence from a supervised administration of an instrument on their published accepted list. No society accepts an unsupervised online result, and no certificate substitutes for either route.

The detail that catches most applicants out is that accepted-test lists change. Instruments are removed as their norms age, and tests that qualified people a decade ago frequently no longer do. Always check the society's current list rather than a summary written elsewhere, including this one.

1 Why Thresholds Are Percentiles, Not Scores

Every society states admission as a percentile. Mensa's own materials describe the requirement as scoring at or above the 98th percentile on an approved test, and the equivalent scores are then listed per instrument rather than as a single number.

The reason is a technical problem covered in IQ Score vs Percentile. Not every instrument uses the same standard deviation. The Wechsler scales use 15, older Stanford-Binet forms used 16, and Cattell scales used 24. The same numeric score therefore maps to different percentiles depending on which test produced it.

Consider what a bare score threshold would do. A rule admitting anybody scoring 132 or above would admit the top 2 percent of Wechsler takers and roughly the top 9 percent of Cattell takers, from the same two digits. The society would end up with two different populations under what looked like one rule, and the rule would be doing something nobody intended.

Stating the threshold as a percentile removes the ambiguity entirely. The 98th percentile is the 98th percentile regardless of instrument, and each accepted test's qualifying score is then derived from its own norms. This is why society published tables list a different qualifying number for each accepted instrument rather than one universal figure.

It also means any conversation about whether a particular score qualifies is unanswerable without knowing which test produced it. Somebody reporting a score of 132 has said nothing about their percentile until they name the instrument, and this is the single most common source of confusion in discussions about eligibility.

2 The Thresholds Compared

Societies are distinguished almost entirely by how selective they are. The percentile determines everything else about them, including how many members they can possibly have.

The final column is an approximation derived from the normal distribution assuming a standard deviation of 15, and it is presented to make the thresholds legible rather than as a qualifying figure. The actual qualifying score for any given instrument comes from that instrument's norms, and each society publishes its own table.

Two things are visible immediately. The step from Mensa to Intertel doubles the selectivity while moving the score by only about five points, which is the percentile compression described in section 1. And the step from the 99.9th percentile upward moves into territory where, as section 6 explains, the measurement itself becomes the problem.

Membership numbers follow directly. Mensa has tens of thousands of members worldwide because the 98th percentile describes a large group. The societies at the extreme end have memberships measured in dozens, not because they are exclusive by policy but because the population that could qualify is small and only a fraction of it applies.

It is worth pausing on how large the qualifying population actually is at the lower thresholds, because the framing of these organisations obscures it. The 98th percentile describes roughly one person in fifty. In a city of a million adults that is twenty thousand people, in a large secondary school it is several pupils per year group, and in any sizeable workplace it is a handful of colleagues. The threshold is genuinely selective in the sense that most people do not clear it, and it is nowhere near as rare as the surrounding language implies. Anybody who has worked in a technical field or attended a selective university has been surrounded by people who would qualify, most of whom have never considered applying.

3 The Two Admission Routes

Every society operates the same two routes, and choosing between them is mostly a question of what evidence you already hold.

The society's own supervised test. Mensa administers a supervised admission test through its national organisations, at scheduled sessions and for a modest fee. This is the route most applicants take, because it requires no prior testing and the result is definitionally acceptable to the body administering it. Sessions run at fixed times and locations, and availability varies considerably by country.

The smaller and more selective societies frequently do not administer their own test, relying instead on accepted prior evidence and in some cases on their own unsupervised high-range instruments, which is a specific and contested case discussed in section 6.

Prior evidence. A score from a supervised administration of an instrument on the society's accepted list, documented by whoever administered it. This is the route for people who have already been assessed, typically through a clinical or educational evaluation, and it requires the documentation described in IQ Test Certificates: a report naming the instrument, the date, the score, and the examiner.

Two practical notes on the prior evidence route. The score must come from a supervised administration, so a school or clinical assessment qualifies while anything self-administered does not. And the instrument must appear on the current list, which is a stricter requirement than most applicants expect, for the reasons in the next section.

A third route exists at some societies for people whose qualifying evidence is unusual, typically involving submission to a membership committee for individual consideration. This is not a shortcut and generally applies to older or non-standard instruments rather than to weaker evidence.

Which route to choose comes down to a straightforward calculation. If you already hold a supervised clinical or educational assessment on the current accepted list, the prior evidence route costs only the submission and involves no risk of a failed attempt. If you do not, arranging a clinical assessment purely to qualify is a very expensive way to reach an outcome the society's own test reaches for a small fee, and the admission test is clearly correct. The clinical route makes sense only when you wanted the assessment anyway for one of the reasons set out in The Adult IQ Testing Process, in which case qualification is a by-product rather than the purpose.

4 Accepted Tests, and Why the Lists Keep Shrinking

Published accepted-test lists are long, and they are shorter than they used to be. Understanding why prevents the most common disappointment in this process.

Instruments are removed when their norms age past the point of usefulness. A test standardised decades ago compares a modern applicant against a population that no longer exists, and because population performance drifts, the old norms overstate current takers. Admitting on obsolete norms would mean admitting people who do not meet the current threshold, so societies prune.

The clearest example is the treatment of admissions tests. American Mensa accepted the SAT as qualifying evidence historically and stopped accepting scores from tests taken after September 1994, and similarly stopped accepting the GRE after 2001. In both cases the reason was that the test was restructured such that the relationship between its scores and general cognitive ability, which had made it usable as evidence, no longer held in the same way.

That example generalises. A test is acceptable to a society because a defensible mapping exists between its scores and the population percentile. Anything that breaks the mapping, whether renorming, restructuring, or the simple passage of time, removes the test from the list regardless of how good it was.

The practical consequences are three. Check the society's current list rather than any secondary source, including this page. If your evidence comes from an older instrument, check whether that edition specifically is still accepted, since editions are listed separately. And if you were assessed long ago, expect that a fresh supervised administration may be required regardless of what your old score was.

The related question of which edition you took, and why it matters, is covered in WAIS-IV vs WAIS-5.

5 Why Supervision Is Non-Negotiable

No society accepts an unsupervised result, and the reason is structural rather than a judgment about online instruments.

A society admitting at a percentile threshold needs that threshold to mean the same thing for every member. Supervision establishes four things that no unsupervised administration can: that the person named produced the score, that they worked alone, that they had no reference materials, and that this was not the best of several attempts.

None of those four is a property of the test. An excellent instrument administered without supervision fails all of them, and a mediocre one administered under supervision satisfies them. This is why the quality of an online test is not the deciding factor and why a better online test does not become admissible by being better.

Proctored online delivery, using identity verification and camera monitoring, closes most of the gap and some societies have adopted it for their own admission testing. It costs more than unsupervised delivery, which is why consumer testing does not use it, and it is the only route by which a remotely administered result becomes acceptable.

An online score is still useful before applying, and this is the honest framing of its role. Someone whose result sits well below a threshold has learned that applying is unlikely to succeed, at low cost. Someone whose result sits near or above it has a reason to book a supervised session. That is a real function and it is not the same as qualifying.

6 The Problem With Extreme Thresholds

Above roughly the 99.9th percentile, the requirement runs into a measurement problem that the societies concerned are generally open about.

Standardisation samples for major batteries contain a few thousand people. At three standard deviations above the mean the expected count in such a sample is a handful, and beyond that it approaches zero. Scores in that region are produced by extrapolating a fitted curve into territory where nobody was observed, which is a modelling output rather than a measurement.

Instruments also run out of items. A battery whose hardest item is solvable by most people two standard deviations above the mean cannot distinguish somebody at three from somebody at four, because both answer everything correctly. This ceiling effect is why standard clinical batteries do not report scores in the range the most selective societies require.

Societies at those thresholds therefore rely on specialised high-range tests, typically unsupervised, untimed, and consisting of very difficult problems, normed on self-selected samples of people who chose to attempt them. These instruments face two objections that anybody considering them should weigh. Their norms come from a non-representative population, which is the same criticism that disqualifies inflated consumer tests. And being unsupervised and untimed, they measure persistence and available time alongside ability, since a sufficiently determined person can work on a problem for weeks.

There is also a simple arithmetic problem that the thresholds themselves create. A society admitting at the 99.9999th percentile is describing roughly one person in a million, which worldwide is a population of a few thousand adults, most of whom will never encounter the organisation. Validating an instrument against a group that size is not practically possible, because assembling a representative sample of it would require testing a substantial fraction of the world's population first. The threshold is therefore a statement of intent rather than a measurable criterion, and everybody involved knows it.

None of this makes the societies illegitimate, and their members are generally aware of the issue. It does mean a claimed percentile at that level rests on a much weaker evidential base than a Mensa qualification does, and that the gap between the two is larger than the numbers suggest. The general problem of scores at the extreme is covered in Highest IQ Ever.

7 What Applying Actually Involves

For the most common case, applying to Mensa via the supervised test, the process is straightforward and worth setting out.

You contact the national organisation and book a testing session. Availability varies enormously by country and region: densely populated areas may have monthly sessions while others test a few times a year, and some national groups offer supervised testing by appointment where no session is scheduled.

The test itself is supervised, takes under two hours in most implementations, and produces a pass or fail against the threshold rather than a detailed profile. This surprises applicants who expected a report. The admission test exists to answer one question, and it answers it without generating the domain breakdown a clinical assessment or a full battery produces.

Fees are modest compared with clinical assessment, typically in the range of an ordinary consumer purchase rather than a professional service. Membership dues follow annually if you qualify.

Most national groups permit only a limited number of attempts, frequently one, occasionally two. This is a direct consequence of the practice effect discussed in How to Prepare for an IQ Test: repeated attempts raise the score without raising the ability, so unlimited retakes would admit people who do not meet the threshold. Check the retake policy before booking, because a failed attempt may close the route permanently.

If you qualify via prior evidence instead, the process is submitting documentation for review rather than sitting anything, and the review checks the instrument, the edition, the date, the supervision, and the score against the current table.

8 Whether It Is Worth Joining

This is a question of what you want from it, and the honest answer varies more than the marketing on either side suggests.

What membership does provide is a social organisation. Local groups, events, special interest sections, publications, and a network of people who share one specific characteristic. For people who have found it difficult to locate others who enjoy the same kind of conversation, this is a genuine and sometimes substantial benefit, and it is the reason most long-term members stay.

What it does not provide is professional advantage. Membership is not a credential, employers do not weight it, and listing it on a curriculum vitae is more likely to read as unusual than as impressive. Nobody outside the organisation has any obligation to treat it as meaning anything.

Nor does it provide validation that lasts. People frequently report that qualifying felt significant for a short period and then stopped mattering, which is the predictable outcome of seeking external confirmation of an internal question. The membership does not change anything about the person, and the feeling that prompted the application usually survives it.

The clearest way to decide is to ask whether you would attend the events. Membership is participation in an organisation, and somebody who wants the qualification without the organisation is paying dues for a certificate, which puts them back in the territory covered in IQ Test Certificates.

There is one genuinely useful function worth noting. For people who have spent years uncertain whether a persistent sense of being out of step reflected something real, a supervised qualification settles the factual question definitively. That is worth something, and it is worth it once rather than continuously.

9 Where the Societies Came From

The oldest and largest of these organisations was founded in Oxford in 1946 by Roland Berrill, an Australian barrister, and Lancelot Ware, a British scientist and lawyer. Knowing the founding intent explains several features of the current organisations that otherwise look arbitrary.

The original conception was explicitly apolitical and non-hierarchical: a society whose only qualification was performance on an intelligence test, deliberately excluding every other criterion including wealth, background, education, and profession. The name was chosen to evoke a round table, at which members sat as equals.

That founding logic is why the threshold is a single number rather than a set of criteria, and why the organisations have historically resisted attaching any further requirement. It is also why they have no professional function. An organisation defined by excluding every criterion except one cannot then claim the credential value that comes from assessing the others.

The proliferation of more selective societies followed a predictable pattern. If an organisation can be defined by a threshold, a higher threshold defines a different organisation, and successive groups were founded through the second half of the twentieth century at progressively more extreme percentiles. The most selective of these were founded specifically to address the perception that the 98th percentile was insufficiently restrictive.

What that history produces is a landscape where the selectivity gradient is very steep and the measurement quality gradient runs in the opposite direction. The least selective organisation admits on the best evidence, because the 98th percentile is squarely within the range clinical instruments measure well. The most selective admit on the weakest evidence, because nothing measures reliably where they set their bar. That inversion is the single most useful thing to understand about the field, and it is a consequence of the founding logic rather than a criticism of anybody's intentions.

10 What People Get Wrong

Several beliefs about these organisations circulate widely and are worth correcting directly, because each one leads people to make a decision they would not otherwise make.

That membership is a credential. It is not, and section 8 covers why. The organisation assesses one thing and explicitly declines to assess anything else, which is the opposite of what a credential does.

That the members are uniformly impressive. The threshold selects on one narrow characteristic, measured on one occasion, with error attached. It does not select on judgment, knowledge, diligence, or decency, and members are as variable on all of those as any other group. Anybody joining expecting a room full of remarkable people is working from a misunderstanding of what a percentile selects.

That failing the admission test settles anything. A single administration is one estimate with a confidence interval, taken on one day under whatever conditions applied. Somebody a point or two below a threshold is not distinguishable from somebody a point or two above it, and the pass or fail structure of an admission test conceals exactly that.

That the societies claim their thresholds are meaningful boundaries. Most do not. A threshold has to be set somewhere for an organisation to exist, and the 98th percentile is a round number rather than a point at which anything changes. The distribution is continuous and nothing happens at the boundary.

That there is a hierarchy worth climbing. The inversion described in section 9 means moving up the selectivity ladder means moving toward weaker evidence, not stronger achievement. Somebody who qualifies at the 98th percentile on a supervised clinical instrument holds better evidence about themselves than somebody who qualifies at the 99.99th on an unsupervised high-range test.

That joining will change how you feel. It reliably does not, for more than a short period, and the pattern is consistent enough that it is worth saying before somebody spends the fee looking for it.

11 If You Do Not Qualify

Most people who sit an admission test do not pass it, which follows directly from the threshold: the 98th percentile excludes 98 percent of the population by construction. It is worth setting out what that result does and does not mean.

What it means is that on one supervised administration of one instrument on one day, your score fell below a specified point. That is a fact about a measurement rather than a conclusion about you, and the confidence interval on any single score means a near miss is not distinguishable from a near pass.

What it does not mean is that the underlying question has been answered. The threshold is arbitrary, the measurement carries error, and conditions on the day affect the result. Somebody who tested while exhausted or anxious produced an underestimate that nothing in the pass or fail structure records.

The practical options are limited and worth knowing. Most national groups permit at most one retake, sometimes none, so a second attempt may not be available. The prior evidence route remains open if a supervised clinical assessment exists or is arranged, and a clinical assessment produces something more informative than a threshold decision regardless of the outcome: a domain profile with confidence intervals, which tells you where your abilities actually diverge.

The more useful reframing is that the question people bring to these organisations is rarely the one the test answers. Somebody applying because they have felt persistently out of step is asking about fit, belonging, and whether an internal experience corresponds to something real. A percentile threshold is a poor instrument for any of those, it answers only whether one number cleared one bar, and a great many people who clear it report afterward that it addressed considerably less than they expected.

12 Where an ACIS Result Fits

An ACIS result does not qualify anybody for any society, and that limitation follows from section 5 rather than from anything about the instrument.

Administration is unsupervised. Identity is not verified, conditions are not observed, assistance cannot be ruled out, and repeat attempts cannot be excluded. Those four gaps are exactly what society admission requirements exist to close, and no property of the items or the norms addresses them.

What an ACIS result can do is inform the decision to apply. Sitting a supervised admission test costs a fee, a booked session, and in most national groups one of a strictly limited number of attempts. Knowing roughly where you sit before spending an attempt is useful, and a domain profile is more informative for this purpose than a single figure, because society thresholds are set against a composite while individual profiles vary considerably around it.

The report states its limits plainly rather than implying institutional acceptance. Index scores with percentiles and confidence intervals, the reference group named, the norms documented in technical materials, and an explicit statement that administration was unsupervised and that no organisation is obliged to accept the result.

That is the honest scope of what an online battery contributes to this particular question, and it is smaller than the category usually claims.

13 FAQ: High IQ Society Requirements

14 Best Next Step

If you intend to apply, go to the society's own site for the current threshold table and accepted-test list, because those change and no secondary source stays accurate. Then decide between the supervised admission test and prior evidence based on what documentation you already hold.

Before spending one of a limited number of attempts, it is reasonable to establish roughly where you sit. For what the percentile threshold actually means, read IQ Score vs Percentile. For what supervised assessment involves, read The Adult IQ Testing Process. For a six-domain profile now, take the assessment.

15 Sources Behind This Page

Thresholds and accepted-test policies come from the societies' own published criteria. Because those change, treat the society's current page as authoritative over any summary, including this one.

• Mensa International. What is IQ. States the 98th percentile requirement and explains why admission is defined at a percentile rather than a score.
• American Mensa. Qualifying test scores. The current accepted-instrument list with per-test qualifying scores, and the documented cutoff dates after which SAT and GRE scores are no longer accepted.
• American Mensa. Testing and admission. Describes the supervised admission test, session availability, fees, and retake policy.
• Intertel. Membership qualification. States the 99th percentile requirement and the accepted evidence routes.
• Triple Nine Society. Admission requirements. States the 99.9th percentile requirement and the instruments accepted as qualifying evidence.
• Prometheus Society. Membership. States a threshold near the 99.997th percentile and the specialised instruments used to establish it.
• American Educational Research Association, American Psychological Association & National Council on Measurement in Education. Standards for Educational and Psychological Testing. Requirements that norms be current, and the conditions under which a score is interpretable.
• Crawford, J.R., Garthwaite, P.H. & Slick, D.J. (2009). On percentile norms in neuropsychology: proposed reporting standards. The Clinical Neuropsychologist, 23(7), 1173-1195. Why percentile is the defensible currency for comparison across instruments.
• Voncken, L., Albers, C.J. & Timmerman, M.E. (2019). Improving confidence intervals for normed test scores. Behavior Research Methods. Open access. Why norm-sample size collapses the precision of scores at extreme percentiles.
• Calamia, M., Markon, K. & Tranel, D. (2012). Scoring higher the second time around: meta-analyses of practice effects in neuropsychological assessment. The Clinical Neuropsychologist, 26(4), 543-570. The basis for societies limiting the number of admission-test attempts.
• Trahan, L.H., Stuebing, K.K., Fletcher, J.M. & Hiscock, M. (2014). The Flynn effect: a meta-analysis. Psychological Bulletin, 140(5), 1332-1360. Population-level drift, which is why instruments fall off accepted-test lists as their norms age.
• Pearson (2024). WAIS-5, Wechsler Adult Intelligence Scale, Fifth Edition. The current edition of the most commonly accepted clinical instrument, and the reported score range it supports.